As best as I can tell that "16,000" is an extrapolation from a 1999 observational study evaluating data from the Arthritis, Rheumatism, and Aging Medical Information System. There were 19 NSAID-related GI bleeding deaths.

Those 19 deaths were used to calculate the NSAID mortality rate in people with arthritis - and then used with an estimate of the number of people in the US with arthritis.

Using that chain of logic to go from an observed 19 deaths to 16,000 in the general population is tenuous - and I would add that it has *nothing* to do with the mortality risk associated with taking NSAIDs for anything *but* chronic pain and inflammation.

If you have another way to back up your numbers, I'd be interested to hear it.